1. Field of the Invention
The present invention is directed to optical illumination systems and more specifically to optical illumination systems that incorporate time-modulated light sources and recombining modulators to increase brightness. Moreover, the invention also discloses novel methods for intensifying illumination brightness by time-superimposing multiple pulsed light sources.
2. Description of the Related Art
It is well known in the art that the brightness of a light source cannot be increased by a passive optical system. Here "brightness" is used in the technical sense of optical power per unit emission etendue, where emission etendue is the product of solid angle in the emitted direction times source area measured in a cross-section perpendicular to emitted direction.
Just as a non-attenuating optical system must preserve source brightness, so it must also preserve as invariant the product of the solid angle of the illuminating light and the cross-section area of the focused illumination beam. The divergence (or convergence) of the beam can be decreased if minimum beam diameter is allowed to increase. Conversely, minimum beam diameter can be decreased i the beam is made more divergent (or convergent). However, it is only possible to make both improvements simultaneously if part of the beam is blocked, which reduces collected power. For given fixed source brightness, the light received by an illuminated object of fixed area is thus determined by the solid angle that the illuminating light occupies. A geometrically equivalent statement is that, for given fixed source brightness, the optical power projected by illuminating optics of fixed diameter is determined by the solid angle into which the optics project the light. The optical system designer must ensure that the source is powerful enough to radiate with this fixed brightness into all regions within the lens diameter, and into all directions within the output solid angle, and that the diameter and solid angle are well chosen according to the various constraints of the application. However, once the source brightness, diameter, and solid angle are fixed in this way, the designer can only increase the delivered power by minimizing absorption and scatter within the system; he cannot redesign the system to concentrate more power into the limited diameter and solid angle.
These constraints of fundamental physics significantly limit the optical designer's freedom to increase illumination intensity. For example, the following equation shows that even if an illumination system can collect all rays emitted by a source of width S, maximum possible delivered power is achieved once one chooses the source large enough that .vertline.S.vertline..gtoreq..vertline.P .alpha..vertline., where P is the lens diameter and .alpha. the angle that the optical system projects into. When this condition is satisfied, the lens aperture is completely filled by the source, and maximum intensity is delivered within the projection angle .alpha.. Unfortunately, if S is increased beyond the point needed to fill the lens, the overfilling light cannot be collected, and the extra light that is output by the larger source is therefore wasted. On the other hand, when .vertline.S.vertline.&lt;.vertline.P .alpha..vertline., the source is too small to fill the aperture, and illumination intensity can be increased by increasing the source size, which, for a fixed class of light source, means increasing the power consumption of the source. Image intensity is said to be power-limited in this case.
However, once .vertline.S.vertline.&gt;.vertline.P .alpha..vertline., further increases in source power do not increase image intensity because the additional source area is not collected within the lens aperture. Image intensity in this case is said to be brightness-limited. Loosely speaking, one might say that when the source is brightness-limited, image intensity can only be increased by increasing the brightness of the collected rays; increasing the size of the emitting region to produce `more rays` does not help.
The field size or angle .alpha. is often fixed by the application. To increase image intensity once the brightness limit is reached, the designer can increase the lens diameter P (or equivalently, increase the numerical aperture [NA], defined essentially as the ratio of Lens aperture radius P/2 to object distance). However, technical constraints on lens performance and/or practical constraints on cost often limit the feasibility of increasing the lens diameter. This is particularly true in projection optical systems, where the illuminated object is re-imaged by a projection lens. High quality projection lenses must not only be designed to capture the full angular and spatial extent of the light that is reflected or transmitted by the illuminated object, they must also project a high resolution image of the object using this light. Image aberrations increase as lens diameter is scaled up. Resolution requirements are particularly stringent in photolithography systems. In projection displays the optics frequently include elements for color and polarization separation/recombination whose cost scales very unfavorably with NA. Thus, in photolithography projectors or projection displays it is not easy to increase the NA of the projection system.
Of course, one requirement for maximizing brightness is that the brightest available source be chosen for the system, which essentially means using the source that produces the greatest intensity on each collected ray. It is common practice to use arc lamps in applications that demand high intensity within a limited NA or object size. It is well known that arc lamps are the brightest light sources available, with the important exception of laser sources. From the point of view of geometrical optics, a laser can be considered to be a froze point source, i.e. a source having infinitesimal extent, so that optical systems using laser sources are always power-limited and never brightness limited. Practical issues with laser sources are often cost and size, particularly as power levels rise into the 1-Watt regime and above. Compact arc lamps in the 1000-Watt range can cost several hundred dollars and might occupy .about.200 cubic inches in the illuminator (plus remote power supply).
Depending on the lamp, the portion of the consumed power radiated as visible light might be 200 Watts. The cost of a laser in the 200-Watt range might be tens or hundreds of times that of the lamp, and the laser might occupy tens or hundreds of times the volume. Though the situation may change in the future, for many applications laser sources are often severely underpowered when practical constraints are enforced on cost and size. On the other hand, while practical non-laser sources can provide very high power, they do so from an extended emitting region, which means that in many applications the power they actually deliver does not reach ideal levels before a brightness-limited regime is reached.
What is needed is a way to increase the brightness of the emitting region itself. However, commercial high brightness light sources are usually engineered to generate as much energy within the emission volume as is technologically possible. For example, when an arc lamp is steadily powered above its rated level, its lifetime decreases catastrophically (i.e. drooping from hundreds or thousands of hours to a few hours). Steady output at increased power requires that the lamp must have a larger arc gap; this means that the source is increased in size but not in brightness.
Brightness can often be increased for brief intervals, but the application must permit the increased emission to accomplish its purpose before damage mechanisms in the source are initiated by the accelerated operation. The source must then be switched off for a sufficient interval to hold time-averaged power below the maximum rated level. It is known in the art that total visible light emission can be improved by pulsing a metal halide lamp, even though total power consumption is held fixed in the time-average, but for simplicity we will assume that time-averaged visible light output is neither increased nor decreased by pulsing. For example, in color-sequential displays it is known that one can periodically pulse LED sources in a way that holds their time-averaged emitted power within tolerance, but which alternates periods of intense emission with periods of non-emission in which the display can be reset for the next color or color bit. Similarly, in photolithography it is known that if one keeps lamp emission very low during periods in which the shutter is closed (for example, while silicon wafers are being loaded, aligned, or stepped to the next chip exposure position), one can cycle he intensity to a higher than normal level during the actual expose period (duration usually &lt;1 second). However, this is not useful in applications where light is required continuously.
Another approach that can provide limited increases in brightness has to do with a simplification made in the above discussion of source brightness in an optical system. The emitting region of a source does not usually have uniform brightness or sharply defined edges. For example, the emitting region of an arc lamp is roughly defined by the gap between discharge electrodes (perhaps .about.2 mm), but source brightness is usually highest near the electrodes and falls off in the middle of the gap (as well as decreasing radially outward). The lamp reflector often increases this brightness non-uniformity. In a real system there is usually not a sharp transition between the power-limited and brightness-limited regimes, making it sometimes preferable for the designer to choose a source large enough that some of the dimmer light in the outer regions of the source is not collected, in order to increase the size of the central high brightness region. Thus, the decision about how to choose a source which best matches into the optics involves a tradeoff between efficiency and total collected light. However, as source size is increased his tradeoff becomes increasingly unproductive until a purely brightness-limited regime is reached.
Other techniques for intensity increase have to do with combining two sources, or combining two images of a single source. As per the above discussion, there is a fundamental physical limitation that reduces the benefit attainable from such combination techniques. It is impossible to merge two incoherent rays that propagate from different points or in different directions into a single ray with twice the energy, unless the initial pair have different wavelengths, or are in different polarization states. A limited exception arises if one of the rays is generated by a source that is not opaque. However, in practice this possibility proves difficult to exploit; for example recombination of a polarization-converted beam with the unconverted component by re-imaging it through the arc is typically not found to be very efficient. Thus, two sources that are unpolarized or of matched polarization can only be combined into a single effective source if the combined source is made twice as large, or is made to radiate into twice as large an angle [or some combination thereof]. Such a doubling of beam width or directional divergence is not useful in a brightness-limited situation.
If an unpolarized source is to be used in an application requiring polarized light, the designer can arrange to separate the unpolarized source beam into separate beams or opposite polarization, and can then convert the polarization of one of the beams to match that of the other, both beams thereby emerging In the polarization needed for the application. This effectively doubles the source power in the desired polarization. However, for fundamental reasons it is Impossible to merge the two beams into a common beam of doubled power but unchanged width and divergence. As noted above, in practice there is not a sharp division between the power-limited and brightness limited regimes, and the designer can often arrange for rays from a high brightness region of the converted beam to displace rays from a low brightness region of the other beam. This improvement is not as large as the 2.times. increase that would be obtained if the two sets of rays could actually be merged, but average brightness is increased somewhat.
FIGS. 1a and 1b show a known arrangement for effecting this conversion and recombination. (In a working system the FIG. 1a optics would typically be followed by additional illumination optics, a target, and optics to project an image of the illuminated target.) Light source 100 is of a well-known kind, consisting of an arc lamp 102 with curved reflector 104 (such as a paraboloid) that projects the emitted light as a beam. Light source 100 can alternatively include a lens (not shown) to collimate or focus the output beam. Alternatively, such collimating or focussing functions can be carried out by the reflector 102 alone. Light source 100 projects an unpolarized beam 106 into a polarizing beamsplitter 108, hereinafter referred to as a PBS. Within PBS 108 a polarizing coating 110 divides beam 106 into perpendicularly polarized components; beam 112 with polarization out of the plane of the diagram (S polarization) and beam 114 polarized within the plane of the diagram (P polarization). Mirror 116 folds beam 112 parallel to beam 114, and birefringent element 118 (most commonly a half-wave retarding place) converts beam 112 to P polarization (matching the polarization of beam 114); thus beams 112 and 114 are combined into a wider beam of common polarization. Lens 120 collects much of the polarize light from beams 112 and 114, but in an application that is not power-limited, lens 120 will not be wide enough to collect all the light. Increasing the diameter of lens 120 would require either increasing the NA of the focused double beam, or increasing the size of the illuminated area at focus. Note that light source 100 and lens 120 are not aligned with PBS 108, nor is mirror 116 of the same length as polarizing coating 110, for reasons which may be understood from the simpler layout in FIGS. 1c and 1d. FIG. 1c shows, in schematic form, a plot 150 of the intensity that is present in beam 114 across the diameter of lens 120 if light source 100 and lens 120 are aligned with PBS 108, and if mirror 116 of the FIG. 1a arrangement is removed. The length of dashed lines 122 and 152 in FIGS. 1b and 1d, respectively, represent the diameter of lens 120, which is not wide enough to encompass all of beam 114. FIG. 1b shows in schematic form, a plot 124 of the intensity across lens 120 produced by beams 112 and beam 114 in the FIG. 1a arrangement where light source 100 and lens 120 are not aligned with PBS 108, and where mirror 116 is shorter than the polarizing coating 110. The heights of dashed lines 22 and 152 represent schematically, the average intensity levels (I) of the output beams. In the FIG. 1a arrangement, lens 120 collects less light from beam 114 than in the FIG. 1c arrangement, but the lost light is more than made up for by collection of the high brightness portion of beam 112. The part of beam 114 that is collected in the FIG. 1c arrangement but not collected in the FIG. 1a arrangement has relatively low brightness. This improvement, however, is not as large as he 2.times. increase that would be obtained if beams 112 and 114 could actually be merged into a single (polarized) beam of unchanged width. Some improvement is made by the FIG. 1a arrangement because brightness non-uniformities cause the collection in the FIG. 1c arrangement to be only partly brightness-limited; it is partly power-limited as well. If one tried to accomplish the improved collection of the FIG. 1a arrangement using a source beam that was larger and more powerful (and therefore more completely brightness-limited), the efficiency gain would be less.
The output beam in the FIG. 1a arrangement is polarized. In the case of unpolarized light, the simple arrangement shown in FIGS. 2a and 2b similarly effects a combination of two partly brightness-limited beams, this time from two light sources, 200a and 200b. Non-symmetric alignment of light sources 200a and 200b with mirror 202 creates a brightness distribution, shown in FIG. 2b, similar to that of the FIG. 1a arrangement. In the FIG. 2a arrangement, light beam 204 from light source 200a is directed to the lens 208 simultaneously with light beam 206 from light source 200b which is first folded by mirror 202 and then directed to the lens 208. If light source 200a is not fully brightness-limited in the application of interest, it can be combined with light source 200b in the manner shown, but only at the cost of decreased efficiency, and as light source size is increased such attempts to achieve greater collected power become even less efficient. In fact, the FIG. 2a arrangement is not very practical since the limited increase in collected power that it would provide is much the same as would be obtained by simply using a single light source of larger power. Moreover, a light source of larger power often means a light source of increased emission volume that will provide little or no increase in the power collectable by near-brightness-limited optics. For the same reason, the FIG. 1a arrangement may not actually provide a great deal more light than the simple FIG. 1c arrangement; its advantage is often that it allows a similar amount of light to be obtained with a smaller light source. When such a smaller light source is used in the FIG. 1c arrangement it is significantly power-limited; with a larger light source that is near brightness-limited, the FIG. 1a arrangement typically provides only a modest increase in delivered power.
FIGS. 2c and 2d present an even starker illustration of these brightness constraints. If surface 212 of PBS 210 is a PBS coating, beam 218 will be unpolarized, and cannot have any higher intensity, shown in FIG. 2d, than would be provided by light source 200a alone (with no PBS 210 ), or by light source 200b alone if surface 212 were a mirror.
What is needed is a way to collect as much power from two (or more) brightness-limited or near brightness-limited beams as would be obtained were it possible to combine two such beams into a single beam of unincreased width and angular divergence; more generally, what is need is to project more light from a fixed source volume into a fixed range of directions than is permitted by the operating power limits of available compact sources.